Gen Sale Forecast:A Hybrid CVAE-BiLSTM Framework for Probabilistic Retail Sales Forecasting
International Journal of Drug Delivery Technology · 02.06.2026 · 16 (44s)
DOI: 10.25258/ijddt.16.44s.53 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Drug Delivery Technology, online erschienen am 02.06.2026. Verfasst von Riya Riya und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Retail sales forecasting is associated with the following difficulties: lack of data, stochastic demand, and inadequate uncertainty estimation. In this work, we present GenSaleForecast — a novel hybrid model, which consists of a pre-trained Conditional Variational Autoencoder (CVAE) integrated with a BiLSTM-Attention model and a probabilistic prediction head. CVAE helps to encode a latent space representing demand and allows for better modeling of temporal dynamics, and the probabilistic head allows for uncertainty estimation. Our model outperforms baselines BiLSTM and LSTM by achieving RMSE=2866.59 and MAE=1860.07 on the Walmart Retail Sales dataset compared to BiLSTM (RMSE=3062.42) and LSTM (RMSE=3525.83).
Zitieren
APA
Riya, R., & Yadav, N. (2026). Gen Sale Forecast:A Hybrid CVAE-BiLSTM Framework for Probabilistic Retail Sales Forecasting. International Journal of Drug Delivery Technology, 16(44s). https://doi.org/10.25258/ijddt.16.44s.53
Vancouver
Riya R, Yadav N. Gen Sale Forecast:A Hybrid CVAE-BiLSTM Framework for Probabilistic Retail Sales Forecasting. International Journal of Drug Delivery Technology. 2026;16(44s). doi:10.25258/ijddt.16.44s.53
BibTeX
@article{Riya2026,
author = {Riya, Riya and Yadav, Neha},
title = {Gen Sale Forecast:A Hybrid CVAE-BiLSTM Framework for Probabilistic Retail Sales Forecasting},
journal = {International Journal of Drug Delivery Technology},
year = {2026},
volume = {16},
number = {44s},
doi = {10.25258/ijddt.16.44s.53}
}
EUVIMED